Lung cancer is a common and often aggressive form of cancer. As it is difficult for doctors to detect it early on, people with lung cancer need to receive the best, most targeted therapy in order to make a positive outlook more likely. Immunotherapy is an option, but how can doctors know who will benefit?

doctors discussing lung cancer scansShare on Pinterest
A new predictive model can determine which people with lung cancer will respond to immunotherapy.

According to the National Cancer Institute, lung and bronchus cancer is the second most widespread type of cancer among people in the United States, accounting for 12.9% of all new cancer cases.

This form of cancer often has no noticeable symptoms in its early stages, which can mean that doctors are unable to detect it at first. This means that the outlook following treatment may not be as good as for other forms of cancer.

To ensure the most favorable outcomes for people with lung cancer, healthcare professionals must choose the best type of treatment for each individual. This, however, can be tricky, since it is often hard to tell which person will benefit the most from a particular treatment.

It can also be difficult for a doctor to determine how beneficial newer types of treatments, such as immunotherapy, will be for an individual. Unlike chemotherapy, which involves using specific drugs to attack and destroy cancer cells, immunotherapy works by boosting a person's immune response against cancer tumors.

Now, a team led by researchers from Case Western Reserve University in Cleveland, OH — in collaboration with scientists from six other institutions — has developed a new artificial intelligence (AI) model. The model allows healthcare practitioners to find which people with lung cancer would benefit the most from immunotherapy.

The investigators explain their method and report their findings in a study paper that features in the journal Cancer Immunology Research.

"Even though immunotherapy has changed the entire ecosystem of cancer," explains study co-author Anant Madabhushi, "it also remains extremely expensive — about $200,000 per patient, per year.

"That's part of the financial toxicity that comes along with cancer and results in about 42% of all newly diagnosed cancer patients losing their life savings within a year of diagnosis," he adds. Madabhushi also notes that the new tool he and his colleagues are working on may help doctors and patients decide which therapy suits them best and avoid unnecessary expenses.

New model can predict outcome

Madabhushi explains that he and his colleagues developed their new model based on recent findings that identified the signs that show which cancerous tumors are responding to treatment.

In a previous study, the investigators found that while doctors have typically thought that tumor size was a good indicator of whether or not a therapeutic approach is working, looking at this characteristic alone can be deceptive.

Instead, says Madabhushi, "[w]e have found that textural change is a better predictor of whether the therapy is working."

"Sometimes, for example, the nodule may appear larger after therapy because of another reason, say a broken vessel inside the tumor — but the therapy is actually working," he explains. "Now, we have a way of knowing that."

To develop the new AI model, the team first used data from computed tomography (CT) scans from 50 people with lung cancer. This allowed them to set up a mathematical method able to identify any changes in size and texture taking place in the tumor after exposure to two to three cycles of immunotherapy.

The method found patterns indicating that particular changes in tumors were associated with a positive response to the immunotherapy treatment, as well as with higher patient survival rates.

This study highlighted once again that those lung cancer tumors that show the most noticeable changes in texture are also the ones who best respond to immunotherapy.

"This is a demonstration of the fundamental value of the program, that our machine-learning model could predict response in patients treated with different immune checkpoint inhibitors. We are dealing with a fundamental biological principle."

Study co-author Prateek Prasanna

Earlier this year, co-author Prateek Prasanna received an American Society of Clinical Oncology 2019 Conquer Cancer Foundation Merit Award for research associated with this study.

Going forward, the team is planning to further test their AI method on more CT scans from other sites, and from people treated with different immunotherapy agents.